Accommodation Coefficients of Hydrogen, Helium, Air, and Argon for Chrome Surfaces at Reduced Pressures R A L P H E. PECK AND JOSEPH D. LOKAY’ Department of Chemical Engineering, I l l i n o i s Institute of Technology, Chicago,
T h e purpose of t h i s investigation is to measure t h e accommodation coefficients of chrome surfaces for g a s e s a t reduced pressures. An apparent thermal conductivity is determined for various reduced pressures, from which the accommodation coefficient and the true thermal conductivity can b e calculated. In 1942 Boelter (1) and a s s o c i a t e s found a scarcity of accurate data on t h e thermal conductivity of g a s e s Among 19 reports on t h e thermal conductivity of air, ‘the deviation from t h e mean value w a s as much as 7%. T h e apparent thermal conductivity w a s measured and accommodation coefficients and true thermal conductivities of t h e g a s were calculated from t h e s e data. Applying t h e kinetic theory of g a s e s to the concept of thermal conductivity l e a d s to t h e conclusion that the heat conductivity should b e proportional to t h e product of t h e specific heat and t h e viscosity. As t h e coefficient of viscosity and the specific heat a r e substantially independent of t h e pressure of the g a s i n t h e low pressure range, ,thermal conductivity should also b e independent of t h e pressure. Experiments performed by Stefan (12) confirmed that t h e thermal conductivity of a g a s is independent of pressure at a range of 428 to 760 mm. of mercury. Kundt and Warburg (6) showed independence for the thermal conductivity of a i r at pressures of 0.5 to 19.5 mm. of mercury. All t h e many subsequent investigators have experimentally confirmed the law. However, Kundt and Warburg (6) first noticed that the thermal conductivity varied with pressure a t pressures lower than 0.5 mm. of mercury. Quantitative measurements of t h i s variation were made by Brush (2) and Smoluchowski (9). Careful experimental measurements show that at very low pressures the thermal conductivity becomes proportional t o pressure, decreasing linearly with it (33. In the pressure range over which thermal conductivity shows t h i s variation, the mean free path of t h e g a s molecules is of the same order of magnitude as the dimensions of the system. In t h i s range the mechanism of energy transfer between the heat transfer surfaces and the rebounding molecules (accommodation coefficient) becomes important, depending not only on the mean free path of t h e molecules and t h e system’s dimensions but a l s o on the physical nature of the heat transfer surfaces. CONDUCTION OF HEAT I N A RAREFIED GAS
T h e theory of heat conduction assumes a special form when the mean f r e e path of the molecule i s larger than t h e dimensions of the apparatus. Molecules then transport heat across the whole apparatus a t a single bound, s o that the whole gas must be assumed to b e a t a uniform temperature. Thus, there is no longer a temperature gradient. A mathematical theory for this case h a s been developed by Knudsen (5). He s t a t e s that the heat transferred through contact ‘ R e s e n t address, Westinghouse Electric Corp., Commercial Atomic Power Activities, Pittsburgh, Pa. 1958
Ill.
with a rarefied g a s will b e jointly proportional to the temperature difference and t o the pressure. Experiments have confirmed the proportionality of t h i s theory, but not the multiplying factor i n the formula. T h e actual heat transferred w a s found to b e less than that predicted. Smoluchowski (10) in 1898 suggested, a s a n explanation, that the exchange of energy from the solid surface to the absorbing molecule w a s far from complete. S d d y and Berry ( 1 1 ) a l s o proposed this theory. T h i s led Knudsen (5) to introduce a quantity a which h e described as a “coefficient of accommodation. ” He assumed that, when a molecule strikes the surface, i t s temperature is not adjusted through the whole temperature difference range, but only through a fractional amount, a, of t h i s range. T h e value of a depends very largely on the temperature and other physical conditions, such a s surface conditions, and may vary from 0 to 1. For carbon dioxide contact with platinum heavily coated with platinum black, Knudsen found a t o b e 0.975. Experiments performed by Roberts ( 8 ) illustrated the variance of a with temperature. A study of the relation between a and the temperature suggested that a would vanish a t the absolute zero. J e a n s (4) developed a simple formula for a based on the mass of the g a s molecule and the mass of the surface molecules. Various attempts have been made to develop a formula for a which agrees with experimental values, but none h a s been developed. T h e problem seems to involve forces which c a u s e absorption of the molecules on the surface. T h e theoretical calculations have been based on the supposition that the only energy which the impinging molecule can transfer t o the wall is i t s kinetic energy of motion. Molecules have many other forms of energy, and the question a r i s e s as to what happens to t h i s other energy when the molecule impinges on a solid. Knudsen (5) h a s devised methods for probing t h i s question and h a s reached the conclusion that the internal molecular energy also h a s an accommodation coefficient, which, within the limits of experimental error, is the same a s the accommodation coefficient for the kinetic energy of translation. Experimental determinations of thermal conductivities of g a s e s have been confined to three principal methods: the cooling thermometer method, the hot wire method, and the plate method. T h e latter method was used i n this investigation. DEVELOPMENT OF THEORY
When heat is transferred by conduction between two parallel solid surfaces, there is a discontinuity in the heat transfer process a t the solid interface. A g a s molecule, leaving a solid surface, will have the surface temperature only if it h a s been absorbed on that surface. At pressures where little gas is absorbed, a molecule striking a hot surface may rebound from t h e surface with practically no increase in temperature. C H E M I C A L AND E N G I N E E R I N G D A T A S E R I E S
37
Heat transfer per square foot of solid surface may be expressed a s
where T , is the g a s temperature next t o the solid surface, T a is the solid surface temperature, and ha i s a n accommodation coefficient. l / h a would be the resistance to heat transfer at the surface. Most accommodation coefficients are defined a s :
D
Figure 1. Heat transfer assembly
The coefficient a s used in these results is related to by the equation:
a
1
a=-
k
-+
dha
APPARATUS
1
It is much more convenient to express the results in terms of, ha but values of a can be calculated from the results. T h e total resistance to heat transfer between two parallel surfaces could b e expressed a s :
where l / h , is the resistance a t each surface and d / k is the g a s resistance. k , is the apparent thermal conductivity and is defined by the equation
k d
q =L(Tl
- T,)
where T, and T, are the two surface temperatures, d is the distance between the surface, and q is the heat transferred per hour per square foot. Thus, if l / k l , the experimentally determined apparent thermal conductivity (Equation 2 ) , is plotted against reciprocal of the distance between surfaces l / d for a constant pressure, the intercept would b e l / k and the slope would be 2 / h a (Equation 1 ) . T h e accommodation coefficient v a n e s with the surface composition, the surface conditions, the gas, and the gas pressure. In a plot of l/k, vs. l / d , the slope 2/ha approaches zero as the pressure of the gas increases from 0 to 1000 microns. At pressures higher than 1000 microns, the slope essentially can b e considered zero. Therefore, under normal pressure conditions, the term 2/ha can b e dropped from the heat transfer equation. As the mean free path increases, the accommodation coefficient becomes more important. Finally, when L (mean free path) is larger than the distance between the heat transfer surfaces, collisions between molecules are few and the temperature gradient in the gas disappears. In this case (free molecule conduction), the effect of the accommodation coefficient is great. When L is larger than the distance between surfaces, there are few collisions between gas molecules and there is no mechanism for molecular temperature change other than collision with the surfaces. In this case, there i s no temperature difference in the gas phase and all the resistance i s at the surfaces. At these pressures, the heat flow is almost independent of thermal conductivity of the gas. Where d is much less than L, k approaches infinity as the At across the gas approaches' zero. There is a finite heat transfer and the only temperature jump is a t the surface. At this low heat flux i t is hard t o obtain accurate data but results indicate a large increase i n k.
38
I N D U S T R I A L AND E N G I N E E R I N G C H E M I S T R Y
A complete description of the equipment was given by Lokay (7). T h e main features which were vital t o the results are shown i n Figure 1. C is the electrically heated sandwich, A and B are the guard plate and ring, and D is the cold plate. To determine the distance between the plates, a hardened steel ball bearing was held against each surface and a series of readings of ball surface was taken with a cathetometer. The contour of each surface w a s obtained in this fashion, and the plate spacing was calculated. The plate spacing was measured to the nearest thousandth of an inch. Thermocouples were installed on the surface of the plates by drilling two holes through the back of each plate. A fine groove w a s drilled between the holes and the thermocouples were placed in the groove. T h e residual space was filled with solder, and the plate was resurfaced. The temperature contours were measured for the hot plate, the cold plate, the guard plate, and the guard ring. Temperatures were measured to the nearest 0.05' F. The apparatus shown in Figure 1 w a s placed in a vacuum chamber where the pressure could be reduced to l e s s than IO-' micron. T h e greatest error i n results probably would be caused by heat l o s s to the guard ring or plate. T h e upper plate on the heating sandwich has to be a t a higher temperature than the lower surface. T h i s deviation is caused by heat flux through the bottom copper plate. Thus, it is impossible t o hold the temperature of B , Figure 1, a t the sandwich temperature. T h e top plate of C was always held at the temperature of the guard plate, A, for the experimental results, Some runs were made where temperature a t A was above and below the temperature on top of C, but 0.1' did not affect the results a great deal. Another source of error was convection. T h e apparatus was shielded by g l a s s plate and glass wool which cut down any stray currents that originated in the equipment. The top plate w a s always used a t a higher temperature, which tended t o dampen any thermal convection inside the unit. Calculated thermal conductivity was independent of the temperature difference between the plates. Radiation w a s determined by dropping the pressure to less than 10'' micron. T h e radiation convection should be almost independent of spacing, and this w a s found t o be the case. A i l results were based on the geometric measurements of the apparatus; however, the results for air checked the best results that have been reported in the literature. CALCULATIONS
Determining the apparent thermal conductivity from the experimental data required only calculation of the radiation l o s s e s and measurement of the physical dimensions. Once the apparent thermal conductivities were determined, they were plotted against pressure to give a curve V O L . 3, NO. 1
L. I
N
ILL
Figure 2. Apparent thermal conductivity of helium
VS.
pressure
c
RECIPROCAL
I/a
OF DISTANCE FEET
Figure 3. Calculation of slope and intercept
Air pressure, 5 0 microns Intercept, 1/K,71.3 Slope, 21 ha, 0.928
20
for each plate spacing (see Figure 2 for an example). Values of the apparent thermal conductivity, k,, were then obtained from t h e s e curves for various pressures between 1 and 1000 microns (0.001 and 1.00 mm.). T h e reciprocals of t h e s e apparent thermal conductivities for a constant press u r e and temperature were then plotted against the reciprocals of t h e spacing distance (see Figure 3 for an example). 1958
40 PRESSURE
-
80
60
I
100
MICRONS
Figure 4. Accommodation film coefficient
VS.
prossure
C H E M I C A L AND E N G I N E E R I N G D A T A SERIES
39
Table 1.
Pressure (Mercury), Microns
Calculated Data on Conductivity and Accornrnodotion Coefflclent
Apparent Thermal Conductivity, B.t.u./Hour Foot O F . , at Spacing of
Slope, 2 -
Intercept, -1 k
Accommodation “ F i l m Coefficient,” B.t.u./Hour Foot O F . , ha
Thermal Conductivity, Bt.u./Hour Foot O F . ,
k
0.250 in
0.525 in
0.995 in.
ha
5 10 20 30 40
0.00228 0.00423 0.00610 0.00733 0.00824
0.00407 0.00603 0.00798 0.00933 0.0102
0.00500 0.00683 0.00902 0.0105 0.0115
4.16 1.84 1.26 1.16 0.926
164 127 100 81.3 77.0
0.481 1.09 1.59 1.72 2.16
0.006 10 0.00787 0.0100 0.0123 0.0130
50 60 70
0.00863 0.00908 0.00943 0.00976 0.0101
0.0109 0.0114 0.0118 0.0122 0.0125
0.0122 0.0127 0.0132 0.0135 0.0138
0.928 0.882 0.808 0.780 0.692
71.3 68.2 67.1 65.3 65.3
2.16 2.27 2.48 2.56 2.89
0.0140 0.0147 0.0149 0.0153 0.0153
0.0103 0.0128 0.0142 0.0149 0.0153
0.0127 0.0146 0.0155 0.0157 0.0158
0.0141 0.0156 0.0159 0.0160 0.0160
0.704 0.328 0.164 0.125 0.089
63.2 62.7 61.5 61.2 61.6
2.84 6.10 12.20 16.13 22.4
0.0158 0.0159 0.0163 0.0163 0.0162
For Air
80
90 100 250 500
750 1000
For Helium 0.252in.
0.495h
0.745 in
0.970 in.
40
0.0052 0.0068 0.0089 0.0105 0.0120
0.0090 0.0113 0.0143 0.0170 0.0193
0.0113 0.0155 0,0200 0.0236 0.0268
0.0125 0.0200 0.0273 0.0322 0.0365
3.185 2.640 1.780 1.960 1.485
37.60 21.60 19.00 14.90 11.65
0.63 0.76 1.12 1.02 1.35
0.0266 0.0463 0.0526 0.0670 0.0858
70 90 110 200 400
0.0160 0.0185 0.0213 0.0320 0.0445
0.0253 0.0290 0.0328 0.0415 0.0545
0.0350 0.0405 0.0460 0.0595 0.0690
0.0465 0.0518 0.0573 0.0735 0.0775
1.130 0.962 0.820 0.475 0.358
11.30 11.00 10.70 9.82 10.62
1.78 2.08 2.44 4.21 5.60
0.0885 0.09 10 0.0935 0.1002 0.0941
600 800 lo00 1600 2200
0.0520 0.0568 0.0603 0.0675 0.0710
0.0620 0.0673 0.0715 0.0780 0.0805
0.0730 0.0765 0.0792 0.0828 0.0845
0.0800 0.0820 0.0835 0.0855 0.0870
0.179 0.148 0.133 0.086 0.072
10.54 10.65 10.65 10.70 10.60
11.18 13.50 15.05 23.30 27.80
0.0949 0.0940 0.0948 0.0935 0.0945
2800 3400 4000 4600
0.0735 0.0758 0.0778 0.0795
0.0185 0.0828 0.0840 0.0850
0.0858 0.0870 0.0878 0.0885
0.0883 0.0890 0.0895 0.0895
0.065 0.057 0.050
10.55 10.63 10.67 10.68
30.80 35.10 40.00 50.00
0.0948 0.0940 0.0938 0.0937
5 10 20 30
0.040
For Hydrogen 0.525 in
0.959 in.
0.0060 0.0125 0.0185 0.0240 0.0270
0.0115 0.0235 0.0320 0.0375 0.0415
7.38 3.46 2.11 1.39 1.19
0 0 5.9 9.9 9.8
0.271 0.578 0.948 1.44 1.68
0.169 0.101 0.102
60 70 80 90
0.0285 0.0308 0.0335 0.0355 0.0378
0.0430 0.0452 0.0487 0.0510 0.0536
1.09 0.962 0.866 0.7% 0.725
10.2 10.5 10.1 10.0 9.9
1.83 2.08 2.31 2.51 2.76
0.0980 0.0950 0.099 1 0.100 0.101
100 250 500 750 1000
0.0403 0.0590 0.0730 0.0830 0.0910
0.0565 0.0770 0.0870 0.0950 0.101
0.657 0.361 0.204 0.139 0.102
9.8 8.3 9.0 8.8 8.7
3.04 5.54 9.80 14.4 19.6
0.250 in.
0.525 in.
5 10 20 30
0.00244 0.00370 0.00494 0.00567
0.00297 0.00435 0.00580 0.00664
2 90 1.60 1.20 1.03
270.3 193.4 145.1 126.9
0.69 1.25 1.67 1.94
0.00370 0.00517 0.00689 0.00788
40 50 60
0.00622 0.00664 0.00701
0.00727 0.00777 0.008 18
0.917 0.790 0.810
118.4 107.1 103.8
2.18 2.28 2.47
0.00845 0.00934 0.00963
70 80
0.00732 0.00760 0.00784 0.00810
0.00852 0.00881 0.00903 0.00932
0.763 0.707 0.685 0.643
100.0 98.1 94.4 92.6
2.62 2.83 2.92 3.11
0.0 100 0.0102 0.0 106 0.0108
5 10 20
30
40 50
0.102 0.120 0.111 0.113 0.115
For Argon
90 la3
40
INDUSTRIAL AND E N G I N E E R I N G C H E M I S T R Y
V O L . 3, NO. 1
28
24.
20
~
16.
12
8.
4
PRESSURE
-
MICRONS
F i g u r e 5. Accommodation f i l m coefficient
A s can be seen from t h e derived Equation 2, the slope of this straight line will b e 2/h,, and t h e intercept (infinite spacing) will b e l/k, the reciprocal of the true thermal conductivity. T h e values of h a and k for air, helium, hydrogen, and argon are given in Table I and Figures 4, 5 , and 6. T h e data (7) show that the thermal conductivity decreases with pressure down to where t h e mean free path equals the plate spacings. At this point the data show a rise going off toward infinity. T h e thermal conductivity should go toward infinity for t h i s dimension of apparatus as the temperature gradient i n t h e p s approaches zero. A l l the curves were then extrapolated t o zero when t h e limit of the experimental apparatus was reached. DISCUSSION
There are, unfortunately, no values of the accommodation coefficients i n t h e literature with which t h e v a l u e s given in T a b l e I may b e compared. T h i s quantity is a function both of t h e g a s and the heat transfer surfaces. One general fault with most of the literature concerning accommodation coefficients is that the preparation of the surfaces and the gases is not described. T h i s makes comparison of values very difficult. T h e surface i n t h i s inves-
1958
VS.
pressure
tigation was a chromium-plated surface and w a s cleaned with a jeweler’s rouge polishing cloth. T h e results indicate that helium and hydrogen have an apparent thermal conductivity which is a function of pressure. If t h e s e results are corrected for accommodation, these g a s e s have conductivities which are independent of pressure down to under 500 microns, a s is the case with other gases. CONCLUSIONS
The experimental data confirmed t h e relationship among the apparent thermal conductivity, the true thermal conductivity, and the accommodation coefficient for the pressure range of 10 to 1000 microns of (mercury) pressure. T h e thermal conductivity of g a s e s a t a low reduced pressure (100 microns or less) varies with pressure. T h e conductivity decreases with a decrease in pressure. T h i s relationship is not a linear one except in the very low pressure range (1 micron or less). T h e thermal conductivity is essentially constant for the pressure range of 200 microns (mercury) up t o atmospheric pressure.
CHEMICAL A N D ENGINEERING D A T A SERIES
41
+I )
-I---600 PRESSURE
700
1 1
8(
900
I(
30
- MICRONS
Figure 6. Thermal conductivity of a i r VI. pressure
T h e value of the accommodation coefficient increases with pressure toward a magnitude which c a u s e s i t t o vanish from t h e heat transfer equation for a g a s a t atmospheric pressure. ACKNOWLEDGMENT
It is a pleasure t o acknowledge the support of the investigations presented here by funds from t h e Office of Army Ordnance Research, Durham, N. C., and t h e Shell Companies Foundation, Inc. LIT E RAT UR E C I T E D (1) Boelter and Tribus. “An Investigation of Aircraft Heaters. Properties of Gases,” W-g Wartime Reports, Univ. Calif., ARR, October 1942.
42
INDUSTRIAL AND ENGINEERING C H E M I S T R Y
(2) Brush, C. F., Phil. Mag. 45, 31 (1898). (3) Dickens, A. R., Proc. R o y . SOC. (London) 1934, A143. (4) Jeans, J,, “Introduction to the Kinetic Theory of Gases,” pp. 185-97, Macmillan, New York, 1940. (5) Knudsen, K, Ann. Physik 31, 205 (1910); 33, 1435 (1910); 34, 593 (1911); 36, 871 (1911); 6, 129 (1930). (6) Kundt and Warburg, Ibid., 156, 177 (1875). (7) Lokay, J. D., unpublished Ph.D. thesis, Illinois Institute of Technology, Chicago, I l t , 1955. (8) Roberts, J. K., Proc. R o y SOC.A129, 146 (1930); A135, 192 (1932). (9) Smoluchowski, M., Phil. Mag. 46, 31 (1898). (10) Smoluchowski, M., Wied. Ann. 64, 101 (1898). (11) Soddy, F., Berry, A. J., Proc. Roy SOC. A83, 254 (1910); A M , 576 (1911). (12) Stefan, Wien. Sitzungersberg. XIV, Abth. 2, 45 (1872). (13) Thomas, L., J. Am. Chem. SOC. 43, 1036 (1943); J. Chem. Phys. 18, 1367 (1950). Received for review June 22, 1956. Accepted March 27, 1957.
VOL. 3, NO. 1